delu.improve_reproducibility

delu.improve_reproducibility(base_seed, one_cuda_seed=False)

[DEPRECATED | Instead, use delu.random.seed and manually set flags mentioned in the PyTorch docs on reproducibility] Set seeds and turn off non-deterministic algorithms.

Do everything possible to improve reproducibility for code that relies on global random number generators. See also the note below.

Sets:

1. seeds in random, numpy.random, torch, torch.cuda

2. torch.backends.cudnn.benchmark to False

3. torch.backends.cudnn.deterministic to True

Parameters:

• base_seed (int | None) – the argument for delu.random.seed. If None, a high-quality base seed is generated instead.

• one_cuda_seed (bool) – the argument for delu.random.seed.

Returns:

if base_seed is set to None, the generated base seed is

returned; otherwise, base_seed is returned as is

Return type:

base_seed

Note

If you don’t want to choose the base seed, but still want to have a chance to reproduce things, you can use the following pattern:

print('Seed:', delu.improve_reproducibility(None))

Note

100% reproducibility is not always possible in PyTorch. See this page for details.

Examples

assert delu.improve_reproducibility(0) == 0
seed = delu.improve_reproducibility(None)